Content uploaded by Bence Decsi
Author content
All content in this area was uploaded by Bence Decsi on Jul 09, 2022
Content may be subject to copyright.
Ecological Indicators 141 (2022) 109143
Available online 8 July 2022
1470-160X/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Original Articles
From simple to complex – Comparing four modelling tools for quantifying
hydrologic ecosystem services
Bence Decsi
a
,
c
,
*
, Tam´
as ´
Acs
a
,
c
, Zsolt Jol´
ankai
a
,
c
, M´
at´
e Kriszti´
an Kardos
a
,
c
, L´
aszl´
o Koncsos
a
,
c
,
´
Agnes V´
ari
b
, Zsolt Kozma
a
,
c
a
Budapest University of Technology and Economics, Department of Sanitary and Environmental Engineering, M˝
uegyetem rkp. 3, H-1111 Budapest, Hungary
b
ELKH Centre for Ecological Research, Lendület Ecosystem Services Research Group, 2163 V´
acr´
at´
ot, Alkotm´
any út 2-4, Hungary
c
National Laboratory for Water Science and Water Safety, Budapest University of Technology and Economics, Department of Sanitary and Environmental Engineering,
M˝
uegyetem rkp. 3. H-1111 Budapest, Hungary
ARTICLE INFO
Keywords:
Hydrologic ecosystem services
InVEST
SWAT
Matrix models
Tool comparison
Hotspot analysis
ABSTRACT
The pursuit of good management of our waters poses permanent challenges to the whole society. Decision-
makers often need to dene appropriate and sustainable strategies on interdisciplinary topics, like water man-
agement issues. The rapidly evolving quantication and mapping of hydrologic ecosystem services (HES) is
putting hydrologic and water management issues into an ecosystem services (ES) framework, which can be a step
towards reconciling different aspects of land use and water management. Different tools can be used for
modelling HES, with a wide range according to their basic properties, e.g., structure, methodology, computa-
tional needs, data requirements, reliability, controllability. As a result of that, the numeric values, spatial pat-
terns, and reliability of HES assessments and the uncertainties in their results may differ signicantly.
In this paper, we covered almost the whole palette of HES mapping tools with regards to modelling approach:
we used InVEST, SWAT and two novel rule-based matrix models for the same pilot area, the 1530 km
2
hilly
catchment of the Zala River (Hungary). We mapped three HES: ood control, erosion control and nutrient (total
phosphorus) retention. Our aim was to examine the relevance of the differences between the HES mapping tools
through analysing the spatial differences between the results obtained with the applied. We carried out spatial
similarity tests and hotspot analysis at the computational unit level for the individual HES and in an aggregated
way as well.
As a result of the spatial pattern similarity tests, InVEST and the matrix models showed moderate to strong
correlation (p <0.001) for each HES. Due to that, the novel matrix models could be considered as robust HES
mapping tools on a larger spatial scale (regional or larger). InVEST appeared to be the most efcient HES
mapping tool considering computational demand, result reliability, and data- and expertise requirements. The
results of our study draw attention to the importance and actual shortcomings of the land use and land cover
(LULC) structure in the riparian zone. We pointed out that the studied HES in agricultural areas close to the
watercourse are often disservices (negative HES were provided with the actual LULC scenario compared to a non-
vegetated LULC scenario) due to the nutrient loads from fertilization. We found that parts of the best and worst
HES provisioning areas (hotspots and coldspots) could be delineated without hydrologic modelling, because their
unfavourable combination of environmental factors and LULC conditions themselves determine these areas to be
hotspot or coldspot.
Abbreviations: ES, ecosystem service(s); HES, hydrologic ecosystem service(s); MAES, Mapping and Assessment of Ecosystems and their Services; RBMP, River
Basin Management Plan; MAES-HU, Mapping and Assessment of Ecosystems and their Services in Hungary project; MAES-HU-TWG-H, Technical Working Group -
Hydrology of MAES-HU; InVEST, Integrated Valuation of Ecosystem Services and Tradeoffs model; SWAT, Soil and Water Assessment Tool; GIS-WSB, novel matrix
model HES mapping tool, where soil parametrization was based on the WetSpa distributed hydrological model; GIS-H1D, novel matrix model HES mapping tool,
where soil parameters were from the results of 1D soil hydrologic modelling based on a national soil database; FC, ood control HES; EC, erosion control HES; PR,
total phosphorus retention HES; LULC, land use and land cover; HRU, Hydrologic Response Unit(s), which is the computational unit of the SWAT model..
* Corresponding author.
E-mail addresses: decsi.bence@emk.bme.hu (B. Decsi), acs.tamas@emk.bme.hu (T. ´
Acs), jolankai.zsolt@emk.bme.hu (Z. Jol´
ankai), kardos.mate@emk.bme.hu
(M.K. Kardos), koncsos.laszlo@emk.bme.hu (L. Koncsos), vari.agnes@ecolres.hu (´
A. V´
ari), kozma.zsolt@emk.bme.hu (Z. Kozma).
Contents lists available at ScienceDirect
Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind
https://doi.org/10.1016/j.ecolind.2022.109143
Received 14 March 2022; Received in revised form 29 June 2022; Accepted 1 July 2022
Ecological Indicators 141 (2022) 109143
2
1. Introduction
While the number of ecosystem services (ES) research has increased
signicantly in recent years, the number of studies and publications
dealing specically with regulating ES are lagging behind (Mengist
et al., 2020). Regulating ES are often less recognized than provisioning
services as these services are less clearly related to human wellbeing and
benets to society (Czúcz et al., 2020; Sutherland et al., 2018). At the
same time, there is an increasing pressure on ecosystems, that results in
the loss of these services (Kandziora et al., 2013; Sutherland et al.,
2018): for instance, effects of human activities on land use and land
cover (LULC) conditions are becoming more and more apparent,
sometimes with disastrous results like oods and their consequences
(Lee and Brody, 2018; Rogger et al., 2017; Wheater and Evans, 2009).
While there are other concepts around similar to that of ecosystem
services– like the Nature’s Contribution to people developed within the
IPBES work (Díaz et al., 2018) - the ES framework is especially suitable
for analysing different aspects of pressures, uses and services and putting
the selected elements on one common platform with several options of
valuating these – assessing societal preferences, biophysical quantities,
comparing relative values or sometimes adding monetary values, too
(G´
omez-Baggethun et al., 2016).
Hydrologic ecosystem services (HES) are mostly regulating ES and
are basically connected to ecohydrological processes like carbon,
nutrient and water cycle and energy partitioning (Sun et al., 2017).
There are several essential ES, which are related to HES or the hydro-
logic cycle like carbon sequestration, air quality control, soil generation
(Jin et al., 2018). There is also connection between HES and provi-
sioning ES like timber and food production and the groundwater
dependence of these ecosystems, thus the ES they provide (Jin et al.,
2018; Liang et al., 2021; Pinke et al., 2020). HES are typically less
tangible assets, but their absence can cause monetary and environ-
mental damage (e.g. droughts, oods, inland excess water or water
quality problems) due to their connection to other ES (Kandziora et al.,
2013; Pinke et al., 2020; Sahle et al., 2019). These hydrological pro-
cesses typically have a stochastic and off-site nature, because their im-
pacts occur as a result of probabilistic, non-linear and dynamic processes
spatially away from the ecosystem providing the service (Bai et al.,
2019; Brauman et al., 2007). Studies on HES usually examine a set of
conicting scenarios related to water and water management from a
wide spectrum (e.g. Turkelboom et al., 2021), but they remain often
disciplinary studies (Brauman, 2015). Issues addressed in scenario
building include land use conicts and alternatives, water scarcity
problems, freshwater stress and water management problems (possibly)
caused by climatic extremes like drought, ood and inland excess water
(Boz´
an et al., 2018; Cosgrove and Loucks, 2015; Froese and Schilling,
2019).
The modelling tools usually applied for HES studies rely on a di-
versity of methodological approaches (Lüke and Hack, 2018; Ochoa and
Urbina-Cardona, 2017). From matrix models or spreadsheet models to
semi-distributed or distributed parametric hydrological models there is
a wide range of used tools to quantify HES (Ochoa and Urbina-Cardona,
2017). The most widely utilized tools found in the literature are: Inte-
grated Valuation of Ecosystem Services and Tradeoffs (InVEST) (Sharp
et al., 2018), ARticial Intelligence for Ecosystem Services (ARIES)
(Bagstad et al., 2011), the Land Utilisation and Capability Indicator
(LUCI) (Jackson et al., 2013; previous version called Polyscape) and the
Soil & Water Assessment Tool (SWAT) (Arnold et al., 1998). These tools
differ in major properties, such as unit of calculation, spatial resolution,
temporality, detail and calculation of hydrological processes and model
complexity (Brauman, 2015; Vigerstol and Aukema, 2011). The most
common weakness of some ES modelling tools (especially LULC based
tools and spreadsheet models) in general is the lack of capability for
calibration and validation (Ochoa and Urbina-Cardona, 2017). During
ES mapping, it is essential to provide reliable results based on appro-
priate input data and mathematical description of the hydrologic and
physical processes (Lüke and Hack, 2018; Vigerstol and Aukema, 2011).
It is also necessary to present model uncertainties well (Hamel and
Bryant, 2017; Wang et al., 2018).
HES mapping tools were reviewed, classied and compared with
each other in a couple previous research (Cong et al., 2020; Vigerstol
and Aukema, 2011). However, there were only a few studies in which
the mapping results produced with different tools were compared
spatially and statistically. Most of the studies (e.g. Cong et al., 2020;
Dennedy-Frank et al., 2016; Lüke and Hack, 2018) focused on the
comparison of SWAT and InVEST models. Cong et al. (2020) and
Dennedy-Franky et al. (2016) for instance went further and paid
attention to model calibration and validation and carried out spatial
comparison of the results on sub-catchment level. They highlighted that
despite the results obtained with SWAT and InVEST models showed
signicant correlations, they differed in values, and that the SWAT
model was found to be more accurate or reliable. On the other hand, the
results of the hotspot analysis matched well (Cong et al., 2020).
For ES assessments in general, the so-called matrix model or
spreadsheet approach is also frequently used, which is a very easy-to-
use, quick and unspecic expert-based evaluation tool that can be
applied to any kind of ES (Campagne et al., 2020; Jacobs et al., 2015).
While this is a rather straightforward approach for most terrestrial ES, it
is challenging to apply it in a hydrologic context due to several water-
ow specic issues e.g., the provided HES by a computation unit
could be realized in downslope areas (Brauman, 2015; Grˆ
et-Regamey
et al., 2017; Nedkov and Burkhard, 2012). As a consequence, there is a
lack of matrix models developed for mapping HES (V´
ari et al., 2022).
Since higher tier hydrological models are the dominant tools used for
HES mapping, these require a lot of input data and are labour intensive
(Brauman, 2015; Lüke and Hack, 2018). This poses a major challenge in
many data scarce regions (Schr¨
oter et al., 2015). Due to the fact, that
models are often used without calibration and validation (Ochoa and
Urbina-Cardona, 2017; Sch¨
agner et al., 2013), the spatial upscaling of
HES results modelled with physically sound and sophisticated tools from
well-monitored sites is not yet widespread, so there is need for devel-
oping simpler models and comparing them to higher tier models (Hanna
et al., 2018; V´
ari et al., 2022).
In our study, a comparative HES mapping was performed, using
three types of mapping tools with different properties for the same pilot
area. We mapped three HES: ood control (FC), erosion control (EC) and
phosphorus retention (PR) in this study. From a methodological point of
view, these tools provide almost complete coverage of the scale of
available mapping tools (Brauman, 2015; Ochoa and Urbina-Cardona,
2017). First, we applied two novel, self-developed matrix models both,
utilising the slightly modied results of the indicator development car-
ried out by the Technical Working Group on Hydrology (MAES-HU-
TWG-H) (Kov´
acs-Hosty´
anszki et al. 2019). Explicit hydrological com-
putations are not included in these models, instead, we involved in-
dicators are suitable to associate and approximate the participation and
position of the given computation units in hydrological processes. Next,
we used the InVEST model, which is also a cell-based, temporally static
tool and is capable of calculating some hydrological and physical pro-
cesses (Sharp et al., 2018). Finally, we applied the hydrological response
unit (HRU) based SWAT model, which is a semi-distributed, semi-
empirical, temporally dynamic model used for simulating hydrological
and transport processes in watersheds (Arnold et al., 1998).
The important novelties of our study include (i) the development and
application of novel matrix models, which tools are not widespread
during HES mapping; (ii) the fact, that the similarity between the
mapped HES results with the different tools were evaluated on the nest
possible spatial scale, on computational unit-level; and nally (iii) the
applied technique used for the spatial similarity test between the new
matrix models and of the InVEST model, which was not previously used
in the topic yet, but we considered to be favourable in the context of HES
mapping. Our research had the objectives: (i) to examine the general
applicability of the selected HES mapping tools by analysing the
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
3
coherence of the results they provide, namely how similar the spatial
patterns are for each HES; (ii) to present two novel matrix models and to
evaluate their applicability from the viewpoint of practical use by
comparing them to widely used HES mapping tools; (iii) determine the
overall spatial similarity of the four tools if the three examined HES is
aggregated (iv) to analyse how the tools differ when we focus on
delineating critical or priority areas (hotspots or coldspots) for inter-
vention/restoration or conservation; (v) to examine how similarly the
applied tools describe the effect of main LULC classes per HES; (vi) to
draw the conclusions of the hotspot analysis and highlight the identied
potential land-use based conicts, and also, to reconcile these conclu-
sions with the existing knowledge in the eld of water management and
other disciplines; (vii) based on the above, we wanted to give recom-
mendations to support the decision-making process in selecting the
optimal tool regarding time and data demand, expert knowledge re-
quirements and reliability.
2. Materials and methods
2.1. Study area
To make well-founded conclusions when comparing HES mapping
tools, it is necessary to select a study area, where site-specic measured
data are available. Doing so, the parameter adjustment can be performed
for InVEST, while the calibration and validation procedure can be per-
formed for the SWAT model (Agudelo et al., 2020; Hamel and Bryant,
2017). The results and conclusions of properly calibrated higher tier
models may be used in data-scarce locations with similar environmental
conditions and can even provide a basis for spatial upscaling procedures
(Hanna et al., 2018; V´
ari et al., 2022). Accordingly, we chosen the well
monitored catchment of the Zala River (Hungary, Eastern Europe) -the
largest tributary of Lake Balaton- as study area, where the measured
data (water quantity and water quality) required for the calibra-
tion–validation procedure are available. This watershed was analysed in
several previous studies, most of them focusing on water quality and
quantity or hydrometeorology (Cs´
aki et al., 2020; Decsi et al., 2020;
Hatvani et al., 2020; Jol´
ankai et al., 2020). The mouth of the Zala River
was articially modied in the 1980s to prevent further eutrophication
of Lake Balaton (Hatvani et al., 2020; Herodek, 1986) and some former
wetlands were restored (and named Kis-Balaton Water Protection Sys-
tem (KBWPS)) to provide regulating ecosystem services like erosion
control, nutrient retention and ood control, too (Pomogyi, 1993; T´
atrai
et al., 2000). Since hydrological conditions and water quality in the
KBWPS are inuenced by substantially different drivers than those
dominant in those dominant in other parts of the watershed (Honti et al.,
2020; Istv´
anovics et al., 2007), we only considered the upstream
catchment area of the KBWPS, covering almost 1530 km
2
(Fig. 1).
The LULC conditions of the study area were derived from the Na-
tional Ecosystem Map for Hungary, developed within the Mapping and
Assessment of Ecosystems and their Services in Hungary (MAES-HU)
project in the frame of the implementation of the EU Biodiversity
Strategy to 2020 (European Commission, 2011; Kov´
acs-Hosty´
anszki
et al., 2019; Maes et al., 2013; Tan´
acs et al., 2021). Major LULC (rst
level) categories were dominated by forests (42%) and croplands
including vineyards and orchards (38%), see Figure B.1 (Tan´
acs et al.,
2021). The dominant soils are luvisols in the watershed. In terms of soil
texture, the upper 30 cm consists of loam in over half of the area, while
in the other half 34% is silty loam and 16% is sandy loam (P´
asztor et al.,
2017).
The climate of the study area is moderately cool and moderately
humid. The spatial pattern of long-term average of annual precipitation
shows an increasing trend from east to west (annually 660 mm in the
northeast corner and 800 mm in the western corner). The mean air
temperature follows an inverse trend, being typically 1.0–1.5 degreesC
cooler in the western part than in the north - eastern part of the area
(Dobor et al., 2015; Western Transdanubian Directorate of Water
Management, 2016).
The width of the Zala River’s main channel is 7–20 m, the typical
water depth is 0.5–2.5 m. The riverbed material is mostly sandy and
silty. The slope of the riverbank is relatively steep, ranging between 50
and 75 degrees. We summarize the main water quantity and water
quality parameters of the Zala River (to which we had access): (i) the
Fig. 1. Study area
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
4
annual average discharge at the outlet of the study area was 4.5 m
3
s
−1
,
while the minimum and maximum discharge values were 0.4 to 91.0
m
3
s
−1
in the period between 1977 and 2019, and(ii) we highlighted
basic statistics for the major water quality components at the outlet
cross-section of the study area, covering the period between 1977 and
2012 in Table 1. The mean total nitrogen concentration is slightly, but
under the water quality standards of the country. The mean total
phosphorus slightly does not make the water quality standards of the
country. Due to that, the ecological status of most water bodies did not
reach the good status based on the second River Basin Management Plan
(RBMP) of the study area (according to EU Water Framework Directive
(WFD) classication terms).
Based on the second RBMP of the study area, almost the whole
catchment (99.84%) - including areas downstream of the KBWPS - is
nitrate vulnerable area. Signicant water management problems
affecting the Zala River include point and diffuse nutrient loads and
hydromorphological issues (Western Transdanubian Directorate of
Water Management, 2016).The proposed action plan (with imple-
mentation target of 2021 or 2027) for the control of diffuse pollutants
include: (i) effective nutrient load control at agricultural lands, espe-
cially within riparian zones of watercourses; (ii) erosion mitigation with
grassing over, afforestation and terracing; (iii) reduction of nutrient
loads from animal husbandry (Western Transdanubian Directorate of
Water Management, 2016).
2.2. The quantied HES and the applied tools
We focused on mapping three HES: ood control (FC), erosion con-
trol (EC) and phosphorus retention (PR) in this study. The range of
examined HES was selected on the basis of four approaches: (i) based on
the aforementioned water management related issues of the study area,
these HES have signicant relevance there; (ii) from a practical point of
view, these HES represent a common set of ES that can be quantied by
the applied tools, so each of the four mapping tools was suitable for
mapping these HES, (iii) according to the literature, these are the most
important and signicant HES for human well-being, as they directly
affect the hydrological cycle, the quantitative and qualitative status of
surface- and subsurface water bodies, as well as ood risk (Brauman,
2015; Harrison-Atlas et al., 2016; Keeler et al., 2012), and nally (iv)
these or some of HES were examined in previous HES mapping tool
comparative studies (Cong et al., 2020; Dennedy-Frank et al., 2016;
Lüke and Hack, 2018). FC is described as a regulating ES, as ecosystems
mitigate water ows and prevent the potential damage to economic
assets (Haines-Young and Potschin, 2018; Vallecillo et al., 2020). FC
HES is provided by the ecosystems for reducing peak discharges with
reducing surface runoff, due to that forests, shrublands, grasslands and
wetlands were mentioned as the main providing LULC categories (Val-
lecillo et al., 2020). EC was dened also as a regulating ES (Haines-
Young and Potschin, 2018), where the vegetation prevents soil erosion
during surface runoff and avoids soils to reach the surface water
network, which HES is a major need for soils in Central Europe
(Steinhoff-Knopp et al., 2021). Due to the similarity between driving
processes of FC and EC, positive correlation and synergy was highlighted
between these HES (Hu et al., 2019; Liang et al., 2021). Nutrient,
especially phosphorus retention is also a regulating ES (Haines-Young
and Potschin, 2018), where the nutrient loads are mitigated by the
vegetation via biological, chemical or geomorphic processes within the
landscape, due to that higher efciency of phosphorus retention leads to
better surface water quality (Hopkins et al., 2018).
HES mapping calculations were performed with the following tools
(Table 2): (i) a novel rule-based matrix model, where the soil properties
were dened by the WetSpa distributed hydrological model (Liu and De
Smedt, 2004); (ii) another novel rule-based matrix model, where the soil
properties relied on a national soil database (Mak´
o et al., 2010); (iii)
InVEST ES mapping tool (Sharp et al., 2018) and (iv) SWAT model
(Arnold et al., 1998).
It is important to emphasize that it is not the actual content of the
used data (e.g., which year’s land cover, which period’s hydrometeo-
rological conditions are given) that is decisive, but how the studied tools
lead to a similar spatial and statistical result per HES and in aggregate. In
the case of InVEST and SWAT, the calculated HES were expressed as.
HESact =Xbs −Xact (1)
where HESact is the quantied HES at the cell or HRU level, Xbs is the
modelled average annual amount of HES using bare soil LULC conditions
for the whole catchment and related parameters instead of the actual
LULC conditions, while Xact is the modelled average annual amount of
HES indicator using the actual LULC conditions for the whole catchment
(Tan´
acs et al., 2021) and its parameters. The reason behind this
approach is that it allows the comparison of HES calculated based on
actual and bare soil LULC conditions and their difference infer the
condition of the actual provided HES by the vegetation (as presented in
Eq. (1)). For example, we did not use fertilization loads on arable lands
in the bare soil scenario, but in the case of the model with actual LULC
conditions. Hence, the effects of diffuse nutrient loads on the HES
bundles became observable as disservices too (Decsi et al., 2020;
Shackleton et al., 2016; Zhang et al., 2007).
The detailed calculation procedures, model setups, calibra-
tion–validation processes, the applied objective functions, the param-
eter sensitivity analyses, and further information about modelling tools,
and the interpretation of each HES per tool are detailed in the rst- and
second chapter of the Appendix-A, while the list of applied databases is
highlighted in Table B.1 and Table B.2.
2.2.1. GIS-WSB and GIS-H1D matrix models
For the three examined HES, we aimed to develop compact, easy-to-
use, and easily interpretable indicators that are optimal for the given
process in terms of invested energy demand and expected representa-
tiveness. We computed HES results by combining the matrix model
approach with an adaptation of the Kenessey method, a rainfall-runoff
estimation method developed for the Eastern European region (D’Al-
berto and Lucianetti, 2019). Beside the Kenessey method there are
several static physically based or empirical methods to estimate rainfall
Table 1
Basic statistics of major water quality components at the outlet of the study area
for the interval between 1977 and 2012 (Western Transdanubian Directorate of
Water Management, 2016).
Component Total Nitrogen Total phosphorus Total suspended solids
Dimension [mg l
−1
] [mg l
−1
] [mg l
−1
]
Minimum 0.36 0.01 0.00
Median 3.40 0.28 20.00
Mean 3.75 0.36 43.13
Maximum 20.50 3.27 3530.00
Table 2
Basic properties of the applied tools. Based on: Lüke and Hack, 2018 and own
experience.
Point of comparison GIS-WSB GIS-H1D InVEST SWAT
Spatial computation
unit
Cell Cell Cell HRU
Temporality Static Static Static Dynamic
Temporal resolution None None Monthly/
Annual
Daily
Hydrologic
computation
No No Yes Yes
Data requirements Low Low Medium High
Calibration options Not
possible
Not
possible
Necessary Necessary
Required expertise Low Low Medium High
Time requirement Low Low Medium High
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
5
runoff conditions e.g., curve number, potential runoff coefcient
(Hawkins et al., 2019; Liu and De Smedt, 2004; Mahmoud et al., 2014).
We adopted the Kenessey method for the following reasons (D’Alberto
and Lucianetti, 2019): (i) it is one of the earliest, and thus veried
methods for runoff approximation; (ii) it can be used for spatially
explicit estimations; (iii) it considers the physical characteristics of the
catchment: terrain, soil and land cover conditions; (iv) and it is
commonly used and accepted in Hungary (the GIS-WSB and GIS-H1D
algorithms were originally developed for the mapping of ecosystem
services in Hungary). Kenessey and later other authors provided
empirically derived value sets for the coefcients. This provided us a
sound basis to adopt and modify the original formula to get a qualitative
measure of catchment conditions related to HES.
The Kenessey method assumes that the runoff coefcient (
α
[-]:
volume of event-based runoff divided by the areal precipitation sum) of
a sub-catchment can be estimated based on partial coefcients charac-
terizing the three determinant environmental factors (partial co-
efcients): slope (
α
1), soil type (
α
2) and vegetation cover (
α
3).
According to literature,
α
has a wide range varying from 0.03 to 0.9,
indicating that depending on local conditions, 3–90% of the precipita-
tion leaves the watershed as runoff (D’Alberto and Lucianetti, 2019). We
made three major modications in the usage of the Kenessey approach.
First, instead of calculating the runoff coefcient only for sub-
catchments, we made estimations on the cell-level (distributed param-
eter approach instead of lumped) to explicitly express spatial runoff
patterns. Second, we assumed that not only rainfall-runoff but also
surface runoff-driven erosion and nutrient transport processes can be
evaluated with the method, due to the connection between these pro-
cesses (Gao et al., 2017; Hu et al., 2019; Liang et al., 2021). The
modication of the slope coefcient was based on the logic of the
Topmodel (Beven and Kirkby, 1979), which classies the hydrological
response from different parts of the catchment based on the topographic
wetness index. And nally: the original Kenessey function is additive
(
α
=a1+a2+a3), while we used a multiplicative form (
α
=
a1•a2•a3), because we expected that the product of the three partial
coefcients would stand out better (both positive and negative
combinations):
aHES =aveg,HES •asoil •atopo (2)
where
● aHES is the product, a relative indicator of the mapped HES at the cell-
level [0–1];
● aveg,HES partial coefcient of vegetation (LULC category) [0–1];
● asoil partial coefcient of soil type [0–1];
● atopo partial coefcient characterizing topographic conditions [0–1].
The role of vegetation is not the same for the three examined HES,
therefore different aveg,HES values were assigned to the same LULC
category, depending on the HES in focus. The values of aveg,HES were
chosen according to literature and expert judgement by MAES-HU-TWG-
H (V´
ari et al., 2021). Partial coefcients asoil and atopo express how soil
and topographic conditions inuence the ratio of surface and subsurface
accumulation of rainfall and are proportional to the runoff potential (in
case of cohesive soils or steep slopes the values are close to the
maximum, while coarse-grained soils and gentle slopes mean lower
values, approaching zero).
The two rule-based matrix models differ in the approach of handling
asoil soil factor. In the GIS-WSB we used the potential runoff coefcient
values from the WetSpa distributed hydrological model for the param-
etrization of asoil (Liu and De Smedt, 2004), hence the WSB acronym in
the model name. For the GIS-H1D potential runoff coefcients to esti-
mate asoil factor, based on the outcomes of comprehensive one-
dimensional soil hydrologic modelling using soil hydraulic data from
the Hungarian Detailed Soil Hydrophysical Database, called MARTHA,
which is a national database (Mak´
o et al., 2010). The ground for
developing the GIS-H1D matrix model was that potential runoff coef-
cient values from WetSpa had only a narrow value range in the study
area. While potential runoff coefcient values in WetSPA are based on
soil data from the United States, the database represents local soil con-
ditions appropriately (Liu and De Smedt, 2004). Spatial variability of
soil types was considered in both cases based on the physical soil type of
the topsoil. For this, we used the 0–30 cm USDA texture class map of the
Digital, Optimized, Soil Related Maps and Information in Hungary
(DoSoReMi) database (P´
asztor et al., 2017). See Figure B.2 for further
soil properties of the study area.
With the atopo we aimed to account for both local relief and the
relative position of each cell within the ow accumulation network of
the catchment. We used the reciprocal of the Topographic Wetness Index
(TWI; Beven and Kirkby, 1979) for this partial coefcient, as it is relating
the slope to the upstream contributing area, due to that, the regulating
role of vegetation will be amplied.
2.2.2. InVEST
InVEST is a free and open-source suite of ES mapping modules. It has
separate modules to map and quantify some of the cultural, provision-
ing, regulating- and supporting ecosystem services (Sharp et al., 2018).
InVEST is a cell-based, temporally static model, it is suggested for
decision-making support studies. We used three InVEST modules to map
HES for the study area: We quantied (i) PR using the Nutrient Delivery
Ratio (NDR) module; (ii) EC by Sediment Delivery Ratio (SDR) and (iii)
FC with the Seasonal Water Yield (SWY) module. In the comparative
analysis, we used the results of model calculations performed in a pre-
vious study as input data (Decsi et al., 2020).
2.2.3. SWAT
The SWAT model is a semi-distributed, semi-empirical watershed
model, that is capable to describe rainfall-runoff, sediment runoff and
nutrient runoff for surface and subsurface (divided to baseow and
lateral ow) pathways (Arnold et al., 1998). Sub-basins of the watershed
are split into HRUs, that represent areas with homogeneous land use,
soil type and surface slope (Neitsch et al., 2011). HRUs are the units
where the water balance components are calculated (Arnold et al.,
1998).
The advantage of the model is that it is a process-based hydrological
model, where the runoff is calculated daily, therefore it gives further
possibilities of ES analysis with ner time resolution and detailed
mathematical formulation of hydrological processes. The disadvantage
of the model from the viewpoint of ES analysis is that there is not any
hierarchical structure within a HRU, which creates a barrier in the
analysis of the patterns in water, nutrient, or sediment retention. SWAT
model of the Zala River catchment was built up with almost 3000 HRUs,
where the median HRU area was 9.5 ha.
2.3. Spatial comparison of the applied tools
We analysed how the results and spatial patterns of the different
tools coincided within the same HES and in aggregate too. Studies
comparing HES mapping tools to date have mostly compared vector
les, i.e., values calculated within mostly a sub watershed polygon (sum
or mean) have been compared mostly from SWAT and InVEST, because
the SWAT model performs calculations on the HRU-level (Cong et al.,
2020; Dennedy-Frank et al., 2016). When comparing the results of
SWAT and the cell-based HES mapping tools, we did the same, applying
the Spearman rank correlation to compare two vectors, similar to pre-
vious studies (Cong et al., 2020; Dennedy-Frank et al., 2016).
The novelty of our research is that the developed matrix models
perform calculations on the cell-level, like InVEST, so the results could
be quantied at the cell level. Thus, in examining the similarity between
InVEST and the new matrix models, we compared two raster les.
Different techniques exist for this purpose, basically they could be
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
6
divided into two groups: (i) cell-by-cell comparison methods such as the
Cohen-Kappa statistics; or (ii) procedures that also take into account the
neighbourhood of the cell, such as moving window techniques (Bennett
et al., 2013; Koch et al., 2018; Kuhnert et al., 2005). The moving window
technique was chosen to compare raster les (Kuhnert et al., 2005;
Rocchini et al., 2016). We found it particularly practical to take into
account the neighbourhood of a given cell, as in the case of HES, the
provided ES by a cell has an off-site effect (Brauman, 2015; V´
ari et al.,
2022). Using the method, the downstream and upstream cells of a given
cell can also be considered when comparing two rasters.
So the moving window concept (Rocchini et al., 2016) was applied to
analyse spatial similarity of the cell-based tools (GIS-WSB, GIS-H1D and
InVEST). Second, we performed another similarity test, where the re-
sults of all tools were compared on the HRU level, this was necessary
because SWAT model performs calculations on the HRU-level (instead of
a cell-level). Third, we computed aggregate HES maps for each tool and
used them for further comparison. Fourth, we delineated single and
aggregate HES priority areas, then we analysed the similarity/overlaps
of these areas calculated by the different tools. Finally, we ranked the
major LULC categories in terms of single and aggregate HES provision
(Figure B.3).
For spatial similarity test between the cell-level tools, we used the
moving window concept, Spearman’s Rank Correlation coefcient (
ρ
)
rasters were computed for each tool pair and each HES (Rocchini et al.,
2016). At this time, the HES modelling results from different tools were
compared by determining the
ρ
correlation coefcient in a cell’s
neighbourhood. This algorithm stops at each cell and calculates the
ρ
for
the surrounding cells. The resolution of the raster and the number of
cells that make up the window affect the result of the comparison
(Rocchini et al., 2016; Rudke et al., 2021). Based on preliminary testing
of window size for the spatial correlation analysis (Figure B.5) we chose
to use a 9x9 sized window.
During HRU-level similarity test the results of the GIS-WSB, GIS-H1D
and InVEST tools were averaged according to the polygons of the SWAT
HRUs. After that the Spearman
ρ
correlation coefcients were calculated
between each tool pair. At this point, the HRUs gave the unit of corre-
lation computation, not the cells. Because we compared two vectors
(HES results from two tools), the HRU level spatial similarity analysis
resulted in a correlation matrix (and not in maps).
After quantifying the three examined HES’ spatial similarity, we
wanted to nd and delineate critical areas. Because the examined tools
have different value sets, a scoring system was applied, similar to pre-
viously developed ones for bringing the results from different tools to
uniform value set (Qiu and Turner, 2013; Schulp et al., 2014; Willemen
et al., 2018). Each quantied HES was scored on a scale of one to ten (or
from ‘bad’ to ‘excellent’) at the cell level according to the deciles of its
own values (similar to Schmalz et al., 2016). Then, an aggregated result
for each tool was calculated too: at the cell-level the geometric mean of
the three examined HES scores was computed, by using the geometric
mean, low scores had a greater impact on the aggregated result, which
helps nding the critical areas (Decsi et al., 2020).
To use the HES modelling results for locating and delineating po-
tential restoration sites, we examined the extreme values further. The
results of all mapping tools were examined statistically, and then the
cells belonging to the lower and upper quartiles were delineated. We
examined the proportion of cells in the lower and upper quartiles (for
the worst and the best performing areas) that overlap (i) in a single HES
and (ii) as aggregated score for different tools. In this study, the term of
hotspot is used for the computational units that appear in the lower
quartile of the value set of the actual HES (single or in aggregate), while
the units in the upper quartile will be referred as coldspot.
We examined to what extent these cells (hotspots or coldspots)
coincide for the different tools. Because of this, we produced “a number
of overlapping tool layers map” (with the value set of one to four, where
one means, that there only a standalone tool was indicating hotspot or
coldspot, while four means, that each tool resulted in a hotspot or
coldspot) for each examined single and aggregate HES’ upper and lower
quartile.
Finally, we analysed how LULC conditions affect the single and
aggregated HES results. We determined basic statistics of examined HES
for the different tools classied by LULC. During this, we used the
reclassied HES results with the values set of one to ten. The LULC
analysis was performed on these reclassied maps (Figure B.4). In this
way, we were able to determine which land use contributes the most to
each examined HES in a positive or negative way.
3. Results
Parameter adjustment results for InVEST modules and calibra-
tion–validation and parameter sensitivity analysis results of SWAT are
presented in the second chapter of the Appendix-A.
3.1. Spatial similarity for the cell-based HES mapping tools
As expected, the two matrix models (GIS-WSB and GIS-H1D) showed
the highest spatial similarity for each examined HES (Fig. 2). In the case
of these tools the
ρ
correlation coefcient values were over 0.80 in more
than 75% of the cells. The best correlation resulted for PR (Fig. 3/C).
These results are not surprising since these tools’ methodology is the
same, only the interpretation of the soil parameter was different.
Related to this, there is no major difference in the correlation be-
tween the
ρ
rasters of InVEST compared with the two matrix models. We
got the best match for PR (Fig. 3/C):
ρ
correlation coefcients were
larger than 0.50 in nearly 50% of the cells. For FC, the correlation be-
tween InVEST and matrix models is also positive (Fig. 3/A), but its
strength lags that of PR. In this case, moderate correlation was achieved
in almost 25% of the cells. For EC, the median of correlation coefcient
values was just above zero (Fig. 3/B). More than 50% of the correlation
coefcients were in the range from −0.25 to +0.25. It can be stated that
for EC matrix models and InVEST showed lower similarity of spatial
patterns than for the other two HES.
3.2. Spatial similarity using HRU averages
The HRU-level correlation between InVEST and GIS-WSB and be-
tween GIS-WSB and GIS-H1D tools was stronger than the correlations
between the other tools for each HES, respectively (Fig. 3). InVEST and
the matrix models have moderate to good correlation of all HES except
for EC (in the case of GIS-H1D and InVEST). SWAT showed good cor-
relation with InVEST and GIS-WSB and moderate with GIS-H1D for EC
(Fig. 3/B), while in the case of FC (Fig. 3/A) and PR (Fig. 3/C) there were
only weak to moderate relationships.
Surprisingly, for FC one of the matrix models showed better agree-
ment with SWAT than with InVEST, this can be traced back to the fact,
that the GIS-H1D matrix model’s asoil factors are based on the results of
comprehensive one-dimensional soil hydrologic modelling. So, the soil
factors are described in a more sophisticated and ground way in that
matrix model, than in InVEST, where only curve numbers are dened for
the soil hydrologic groups. It is also interesting that the SWAT model has
good to moderate correlation coefcients for EC, but neither for PR,
while the two processes are physically related to each other. Each cor-
relation coefcient had the signicance level of p <0.001.
3.3. Aggregated results of the different tools
The aggregated maps can be interpreted as a nal score on a scale of
one to ten, in which one shows the areas providing worst for the
examined HES overall, while ten shows the patterns providing the best
HES bundle (Fig. 4).
Similar to our previous spatial analyses, GIS-WSB (Fig. 4/A) and GIS-
H1D (Fig. 4/B) led to a very similar pattern. The aggregate HES map of
InVEST was close to these. However, the number of extremes was less
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
7
(Fig. 4/C). For the aggregated map of the SWAT model (Fig. 4/D the
patterns were in line with the other three tools, but the extremes were
less diverse spatially. The median of the aggregated scores from each
tool were in a narrow interval between 4.76 and 5.24. On the other
hand, the interquartile ranges differed much more. In the case of GIS-
WSB and GIS-H1D this was in the 2.88–7.83 score range, while for
InVEST it was 3.11–7.11 and for SWAT a narrower range with an in-
terval of 3.42–6.60 was scored. Thus, we found that the aggregated
maps show good similarity. From a statistical point of view, the tools
performing hydrological calculations (InVEST and SWAT) led to
spatially less fragmented and diverse results.
3.4. Priority areas of HES (hotspots and coldspots)
Fig. 5 indicates the number of overlapping layers for the lower (reds)
and the upper (blues) quartiles on colour scales (lighter/darker colours
meaning less/more overlapping tools). I.e., the red colours indicate the
areas where single (Fig. 5/A-C) or aggregated (Fig. 5/D) HES are the
weakest. Standalone cells that show a hotspot or coldspot in a cell are
not shown for better overview. Because SWAT had different computa-
tional units for each HES, there is an enormous number of cells which
are presented as hotspot or coldspot by only that tool.
Overall, two overlapping layers category was in the majority for the
HES overlap maps (40%–56%). These came mostly from the agreement
of GIS-WSB and GIS-H1D tools that gave around 50% of this category
(two overlapping layers). The three layers overlap category was between
30% and 39% of the total area. It was the highest in the case of FC
(Fig. 5/A): lower quartile: 38%, upper quartile 39%, while the other two
HES (Fig. 5/B and Fig. 5/C) came out with a ratio near 30% both lower
and upper quartile.
For single HES, PR performed best in the four overlapping layers
category (Fig. 5/C). It meant 24% agreement in the lower quartile, while
26% agreement in the upper quartile. In the case of the other two HES,
the four overlapping layers category’s ratios were in a narrow interval
between 14% and 17%, except for ood control in the lower quartile, it
had only 10% agreement of total coverage.
For the aggregated map (Fig. 5/D) three overlapping layers category
was the dominant in the case of coldspots with 36% of coverage but four
overlapping layers category also reached 30%. For hotspots of the
aggregate map, the two overlapping layers category were in majority
with 44%, while three overlaps reached 30% and four overlaps had 26%
coverage.
Overall, previous experience was further strengthened in the lower
and upper quartiles, suggesting a good relationship between matrix
models and InVEST, while SWAT lagged, making two and three over-
lapping layers class dominant in all cases. Overall, PR showed the best
overlaps, while FC and EC appeared to have weaker agreement. It is
important to point out again the different computational units of the
SWAT model.
Fig. 2. Violin plot of the Spearman’s
ρ
rank correlation coefcients (SRC) of the results maps per tool pairs A: Flood control, B: Erosion control, C: Phosphorus
retention Note: H1D refers to GIS-H1D and WSB refers to GIS-WSB.
Fig. 3. HRU level comparison: Spearman’s
ρ
rank correlation coefcients (SRC) between tool pairs per HES A: Flood control, B: Erosion control, C: Phosphorus
retention Note: *** represents signicance level of p <0.001.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
8
3.5. Effect of land use and land cover conditions on HES values
Finally, we examined the HES results for the major land use cate-
gories per each tool. Fig. 6 shows that the applied tools led to slightly
different results in the urban environment (rst column). This is most
pronounced in the case of EC (Fig. 6/E). Here, InVEST and SWAT rated
these areas signicantly better than the GIS-WSB and GIS-H1D tools, this
can be linked to the parameterization of LULC categories: the matrix
models used vegetation parameters expressed by MAES-HU-TWG-H,
while the other tools were based on literature recommendations and
sample datasets of their model documentation.
Croplands got similar scores in case of each single HES and in
aggregated too. Croplands were judged by all four tools to be particu-
larly weak in terms of PR (Fig. 6/B). Grasslands performed better for
each single HES and in aggregate, than the urban areas and croplands
(column three of Fig. 6). Except for SWAT, the applied tools showed
similar results, SWAT has underrated grasslands for each HES. Forests
turned out to score highest out of all LULC categories overall. SWAT has
turned out to be an outlier with wider value ranges. Interestingly, for
forests GIS-WSB tool slightly gave better scores for each single HES and
in aggregate too, while the other three tools performed quite similarly,
but they always showed a wider value range.
4. Discussion
4.1. Comparison of HES mapping tools
Based on our results, the InVEST model proved to be a good
compromise for analysis. This is in line with the ndings of previous
comparative studies in the topic (Cong et al., 2020; Dennedy-Frank
et al., 2016). It is characterized by medium data demand and moder-
ate labour-intensiveness, and no in-depth expertise is required to use the
tool. Physical-biological background of the described processes are
considered in a simplied way (Harrison et al., 2018; Lüke and Hack,
2017), that proved to be satisfactory for the assessment performed in
this study. At the same time, both the compilation of input map data and
the choice of model parameters allow the utilization of expert knowl-
edge and provide possibility for calibration-verication (Cong et al.,
2020; Decsi et al., 2020; Redhead et al., 2018).
Our own experience underlines the general pros and cons reported
for the SWAT model (Arnold et al., 2012; Tan et al., 2020). As expected,
it provided highly detailed and time-dependent hydrological and water
quality results. In addition, automated calibration, sensitivity testing
and measurement verication of the model is possible and strongly
recommended (Chilkoti et al., 2018; Harrison et al., 2018; Tan et al.,
2020). This poses a major challenge we experienced as the model could
be overparameterized (several different parameter combinations pro-
vide similarly good model ts), making the calibration process labour-
intensive and the results uncertain (Beven, 1993; Seibert et al., 2019;
Whittaker et al., 2010). Another issue we faced in our analysis concerns
HRU delineation. In the SWAT model the ow hierarchy does not appear
explicitly in the computational results and the model generates HRUs
based on only three environmental factors: land use, soil type and slope
(Gassman et al., 2007). Our experience showed that, unrealistic situa-
tions can occur because of these simplications: e.g., hilltops and
riverside valley plains could end up on the same HRU, while it is obvious
Fig. 4. Aggregate scores of the different tools.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
9
that due to the difference in elevation hydrological responses of these
areas would be different. In agreement with previous ndings, we found
that main disadvantages of using SWAT for HES mapping are the
extensive need for data, time, labour, and expertise (Lüke and Hack,
2018). It is important to note that the benets of distributed- or semi-
distributed parameter hydrological models during HES mapping can
only be reaped if the applied models can be compared with site-specic
measured data during the calibration–validation process (Agudelo et al.,
2020; Hamel and Bryant, 2017). This issue signicantly narrows the
scope of the applicability of these higher tier tools, as the lack of the
availability of site-specic measurement data (e.g., water quantity or
water quality) is a common problem and the probability of data scarcity
increases with the extent of the study area (Schr¨
oter et al., 2015). In such
cases the following may be suitable for ES and HES quantication: (i) the
less data-intensive matrix models (Campagne et al., 2020; Harrison
et al., 2018), (ii) the spatial upscaling of hydrological models from the
pilot area level (Hanna et al., 2018; V´
ari et al., 2022), or (iii) the tools
based on articial intelligence techniques or data-driven modelling,
which are gaining ground in ES mapping (Landuyt et al., 2013; Willcock
et al., 2018). Articial intelligence tools have been used successfully to
model and predict water quality in study areas with irregular and small
datasets, both in surface, subsurface, and estuary environments (Hadji-
solomou et al., 2021; Kouadri et al., 2021; Shamshirband et al., 2019).
Nevertheless, application of articial intelligence tools has not yet
gained signicant ground in HES mapping, Bayesian Belief Network
tools have been used primarily (i) to quantify relationships between
different mapped ecosystem services e.g., regulating- and provisioning
services or (ii) to compare the provided ES bundles by the application of
different LULC or production scenarios (Forio et al., 2020; Landuyt et al.,
2016; Pham et al., 2021).
The comparison of simpler- and higher tier models is an important
task to reduce the uncertainties associated with using simpler methods,
and the development of purpose- and region-specic models is necessary
(Forio et al., 2020; V´
ari et al., 2022). The advantage of the introduced
matrix models lies in their simplicity, similarly to previously developed
matrix models (Campagne et al., 2020; Jacobs et al., 2015). This is in
line with the main objective of these tools: to offer a regional or coun-
trywide ES or HES mapping tool that provides easy-to-interpret results,
which may efciently support the regional scale localisation of critical
areas, while requiring low level of data input and work effort (Harrison
et al., 2018; Schlutow and Schr¨
oder, 2021). Disadvantages also come
from simplicity: the developed matrix models use only dimensionless
values for HES mapping. Therefore it is difcult to compare these results
directly to physical processes, quantities, and measurement (Campagne
et al., 2020; Harrison et al., 2018) on the other hand formal calibration is
limited or impossible (Czúcz et al., 2018; V´
ari et al., 2022). In our study,
to eliminate this, we reclassied both physical quantities and dimen-
sionless results from matrix models in their own distributions, making
the matrix methods comparable to the other tools. Another important
shortcoming of matrix models is that the resulting dimensionless values
determined for the level of ES are unlikely to provide sufcient infor-
mation for economic evaluation (Campagne et al., 2020; Harrison et al.,
Fig. 5. Number of overlapping tools in the lower (reddish colours) and upper (bluish colours) quartile of the examined HES.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
10
2018).
Generally, for a set of different ES, matrix models based on structured
expert elicitations showed a rather good t with biophysical models at
the landscape scale (Roche and Campagne, 2019). We found that
InVEST and the two matrix models had moderate to strong correlation
(p <0.001) for FC and PR HES at both computation unit level (cell and
HRU), while in case of EC, results from SWAT and InVEST showed the
third strongest relationship at HRU level (p <0.001). According to the
results of the present study, the results of the SWAT model showed a
signicant but weak correlation (p <0.001) with the novel matrix
models and InVEST for each examined HES, except for the case of EC,
where the signicant correlation was moderate to strong (p <0.001).
Consistent with previous HES mapping tool comparative analyses (Cong
et al., 2020; Dennedy-Frank et al., 2016), we found that the InVEST and
SWAT models showed signicant relationship for each examined HES,
however the HES value sets differed. Being closely related in their
methodological approach, the applied two novel rule-based matrix
models (GIS-WSB and GIS-H1D) led to strong signicant correlations
(with each other) and similar spatial patterns for each HES. Regarding
the novel matrix models, both performed well as HES mapping tools,
and their applicability can be considered as good alternatives of InVEST
and SWAT on a regional scale due to their lower computational and data
requirements. We consider this to be an important result because of the
lack of matrix models developed specically for HES quantication: our
results suggest that the fact, that the mapped processes have offsite
nature (that is HES are not necessarily provided locally), does not
necessarily mean that simple models uncapable to handle such spatiality
should be avoided when mapping HES.
Overall, our ndings support the suggestion, that the application of
simpler tools like matrix models or InVEST can be ideal as the rst step
of an integrated regional landscape development procedure, by delin-
eating critical areas where the aggregate scores of the analysed HES are
the strongest/weakest (Cong et al., 2020; Li et al., 2017; Willemen et al.,
2018). The former are locations prioritised for preservation, while the
latter are potential intervention zones (Decsi et al., 2020). In the second
step the specic scenarios of interventions should be designed with more
sophisticated hydrological modelling (Kozma et al., 2022; Lüke and
Hack, 2018), modelling and evaluation of other concerned disciplines,
eld measurements and analyses within the outlined critical zones. The
involvement of the local government, stakeholders, representatives of
concerned disciplines and risk takers is also important among the
initiating, modelling, designing and constructing steps (Halbe et al.,
2018; Souliotis and Voulvoulis, 2022; Sta´
nczuk-Gałwiaczek et al.,
2018).
4.2. Areal ndings
In the process of identifying priority areas, we the observed, that
some parts of the Zala River catchment were found to be uniformly
critical by multiple tools and could be sharply separated from more
uncertain zones where only one or two tools showed critical values. The
appearance of hotspots and coldspots was clearly dominated by their
environmental factors. At sites with ne-grained soils and higher annual
precipitation sums and higher ratio of forests (e.g., at the western edge
of the study area), the effect of vegetation gets more pronounced and has
a higher HES potential. On the other hand, in the north-eastern part of
the study area, due to high inltration capacity of sandy soils combined
with the more dominant presence of agricultural lands, the need for HES
attenuating water or material ow is less crucial, and the actual vege-
tation adds to less HES capacity too.
Fig. 6. HES scores (rows) by major LULC categories (columns) Note: The colours of the boxplots are dened by the applied tools - yellow: GIS-WSB; blue: GIS-H1D;
red: InVEST; green: SWAT.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
11
Based on our analysis for the three HES, and their aggregate results,
most of the hotspots are arable lands located in the vicinity of the surface
water network. In case of PR, this can be clearly traced back to LULC-
specic nutrient load values (effect of fertilizer nutrient loads in
arable lands). Overall, in line with expectations and previous research
(Anache et al., 2018; Ding et al., 2015), it can be stated, that forests and
grasslands have signicantly better efciencies and capacities than
urban areas and croplands in the range of regulating ES. This nding is
also supported by the LULC category level evaluation of our results, as
agricultural areas received the worst, while forests the best scores
overall. This experience is parallel (i) with relevant international
research (Kardos and Clement, 2020; Lemm et al., 2021; Turunen et al.,
2021) and (ii) with the conclusions of the territorially competent second
RBMP and the preparatory discussion paper of the third RBMP: the
major suggested intervention for improving surface water quality was
the reduction of nutrient loads with LULC change, especially in the
buffer strips of the water network (Western Transdanubian Directorate
of Water Management, 2016). This nding highlights and conrms the
importance of the appropriately chosen LULC conditions of riparian
zones and the need of (semi)natural vegetation buffer strips along water
courses from a HES point of view (Burdon et al., 2020; Haddaway et al.,
2018), in accordance with the WFD and EU Biodiversity Strategy ob-
jectives (European Commission, 2020). These conclusions may provide
a good basis for a spatial upscaling study, where areas with unfavourable
LULC condition near water the network and providing bad HES bundles
could be delineated on a national or larger spatial scale. Such assessment
would be particularly relevant for countries like Hungary, where the
LULC conditions of the riparian zones are the most unfavourable in the
whole Danube River Basin, because the forest to agricultural land ratio is
the lowest in this country. Also, almost 68% of the water courses’ 500 m
wide buffer strip is agricultural land in 2018, which is higher than the
country-level ratio based on LULC conditions (European Environment
Agency, 2017).
5. Conclusions
In this paper, we tried to ll the gap, we have experienced by com-
parison of the lower- and higher tier HES mapping tools. Comparative
studies like ours would be important because they can reduce the un-
certainties of simpler models, in addition well-established simpler
models could be potentially used in data-decient areas if the similarity
of the relevant environmental factors between the area of model
development and the area of potential application allows that. In this
study we compared four tools capable for HES mapping on the Zala
River catchment (Hungary): two novel matrix models (GIS-WSB and
GIS-H1D), the InVEST and the SWAT models. These models differ
signicantly in the modelling approach, temporality, hydrological
descriptive methodology, unit of calculation and resource requirements
including input data, expert knowledge, and time. Therefore, from
practical aspect, it is important to know how reliable the tools are and
whether the simpler ones could substitute the more complex models
when mapping and quantifying HES. For comparison we performed HES
mapping analyses for three HES types on various spatial units (compu-
tational cell-level, HRUs and areas of major LULC categories). Based on
our analysis, the compared HES mapping tools led to similar results in
most examined cases regarding hotspot analysis, LULC-based HES ca-
pacity or aggregated HES mapping.
We primarily recommend using InVEST, a reasonable compromise
(considering description of hydrological processes, computational need,
and reliability of results), as a HES related, supportive decision-making
tool on the watershed scale. The signicant moderate to strong corre-
lations (p <0.001) between the HES results of InVEST and the novel
matrix models suggest that the latter could also be considered as
reasonable alternatives substantially with lower computational effort
but at the expense of information content of the results (no physical
quantities of actual HES are provided). The applicability of simpler
matrix-models for mapping the HES types examined in this study at
watershed scale was not obvious, since the usually offsite nature of these
HES (a phenomena matrix-models cannot handle properly) was the
presumable reason such models were barely developed for HES map-
ping. However, it is to be researched whether matrix-models are appli-
cable for other HES types and at different scales. Due to the statistically
signicant correlation between the results of the novel matrix models
(based on only environmental factors) and the other tools (performing
hydrological calculations), it can be stated for some specic combina-
tion of (favourable or unfavourable) actual LULC conditions and certain
environmental factors, that they can determine themselves the provided
HES.
The shortcoming of our paper is, that we did not examine the envi-
ronmental system in its complexity, but only mapped the ES of a
particular discipline. This is in line with general ES mapping issues, so
synergies and trade-offs between different ES are often not quantied,
making it difcult to incorporate the ndings of ES mapping studies into
the decision-making process. This issue could be solved by an increase in
the number of studies examining multiple ES and their interactions at
the same time and quantifying synergies or trade-offs. We also see the
future of our research in this, it would be expedient to integrate the
research in different disciplines created within the framework of MAES-
HU. This would make existing land use conicts more transparent and
resolvable at the decision-making level. In addition, we would like to
examine in more depth the relationship between critical sites and
environmental factors, thus laying the foundation for a spatial
upscaling.
The consequences of our ndings regarding hotspots within the Zala
River catchment are in line with statements from the eld of water
quality (determined by measured data processing), that the higher
agricultural land use ratio in the riparian zones of watercourses puts
greater pressure on the water quality and ecological status of water-
courses and freshwater ecosystems. This highlights that ES mapping
(with the widest possible discipline coverage) can be a good approach
for detecting and resolving land-use conicts.
CRediT authorship contribution statement
Bence Decsi: Conceptualization, Methodology, Validation, Visuali-
zation, Formal analysis, Writing – original draft. Tam´
as ´
Acs: Concep-
tualization, Supervision. Zsolt Jol´
ankai: Methodology, Validation,
Writing – review & editing. M´
at´
e Kriszti´
an Kardos: Methodology,
Writing – review & editing. L´
aszl´
o Koncsos: Supervision, Funding
acquisition. ´
Agnes V´
ari: Conceptualization, Methodology, Writing –
review & editing, Funding acquisition. Zsolt Kozma: Conceptualization,
Methodology, Writing – review & editing, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
Data will be made available on request.
Acknowledgements
The authors would like to express their gratitude to all members of
the Hydrologic Experts’ Group worked in the Mapping and Assessment
of Ecosystems and their Services in Hungary project: Be´
ata Pataki,
L´
aszl´
o P´
asztor, Zs´
oa Bakacsi, Brigitta T´
oth, Annam´
aria Laborczi, Zsolt
Pinke, G´
eza Jol´
ankai †, Csaba Centeri, Zsolt Matt´
anyi and the other
consortium partners, too.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
12
Funding
The research reported in this paper and carried out at BME has been
supported by the NRDI Fund (TKP2020 IES, Grant No. TKP2020 BME-
IKA-VIZ) based on the charter of bolster issued by the NRDI Ofce
under the auspices of the Ministry for Innovation and Technology and
the European Regional Developmental Funds as part of the Sz´
echenyi
2020, the Environmental and Energy Efciency Operative Programme
and the Competitive Central Hungary Operative Programme (KEHOP-
4.3.0-VEKOP-15-2016-00001). The research presented in the article was
carried out within the framework of the Sz´
echenyi Plan Plus program
with the support of the RRF 2.3.1 21 2022 00008 project.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.ecolind.2022.109143.
References
Agudelo, C.A.R., Bustos, S.L.H., Moreno, C.A.P., 2020. Modeling interactions among
multiple ecosystem services. A critical review. Ecol. Modell. 429, 109103 https://
doi.org/10.1016/j.ecolmodel.2020.109103.
Anache, J.A.A., Flanagan, D.C., Srivastava, A., Wendland, E.C., 2018. Land use and
climate change impacts on runoff and soil erosion at the hillslope scale in the
Brazilian Cerrado. Sci. Total Environ. 622–623, 140–151. https://doi.org/10.1016/j.
scitotenv.2017.11.257.
G. Arnold, J.N. Moriasi, D., W. Gassman, P., C. Abbaspour, K., J. White, M., Srinivasan,
R., Santhi, C., D. Harmel, R., van Griensven, A., W. Van Liew, M., Kannan, N., K. Jha,
M., 2012. SWAT: model use, calibration, and validation. Trans. ASABE 55,
1491–1508. https://doi.org/10.13031/2013.42256.
Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic
modeling and assessment part I: Model developement I. JAWRA J. Am. Water
Resour. Assoc. 34, 73–89. https://doi.org/10.1111/j.1752-1688.1998.tb05961.x.
Bagstad, K.J., Villa, F., Johnson, G.W., Voigt, B., 2011. ARIES–ARticial Intelligence for
Ecosystem Services: a guide to models and data, version 1.0. ARIES Rep. Ser. 1.
Bai, Y., Ochuodho, T.O., Yang, J., 2019. Impact of land use and climate change on water-
related ecosystem services in Kentucky. USA. Ecol. Indic. 102, 51–64. https://doi.
org/10.1016/j.ecolind.2019.01.079.
Bennett, N.D., Croke, B.F.W., Guariso, G., Guillaume, J.H.A., Hamilton, S.H.,
Jakeman, A.J., Marsili-Libelli, S., Newham, L.T.H., Norton, J.P., Perrin, C., Pierce, S.
A., Robson, B., Seppelt, R., Voinov, A.A., Fath, B.D., Andreassian, V., 2013.
Characterising performance of environmental models. Environ. Model. Softw. 40,
1–20. https://doi.org/10.1016/j.envsoft.2012.09.011.
Beven, K., 1993. Prophecy, reality and uncertainty in distributed hydrological modelling.
Adv. Water Resour. 16, 41–51. https://doi.org/10.1016/0309-1708(93)90028-E.
Beven, K.J., Kirkby, M.J., 1979. A physically based, variable contributing area model of
basin hydrology / Un mod`
ele `
a base physique de zone d’appel variable de
l’hydrologie du bassin versant. Hydrol. Sci. Bull. 24, 43–69. https://doi.org/
10.1080/02626667909491834.
Boz´
an, C., Tak´
acs, K., K¨
or¨
osparti, J., Laborczi, A., Túri, N., P´
asztor, L., 2018. Integrated
spatial assessment of inland excess water hazard on the Great Hungarian Plain. L.
Degrad. Dev. 29, 4373–4386. https://doi.org/https://doi.org/10.1002/ldr.3187.
Brauman, K.A., Daily, G.C., Duarte, T.K., Mooney, H.A., 2007. The nature and value of
ecosystem services: an overview highlighting hydrologic services. Annu. Rev.
Environ. Resour. 32, 67–98. https://doi.org/10.1146/annurev.
energy.32.031306.102758.
Brauman, K.A., 2015. Hydrologic ecosystem services: linking ecohydrologic processes to
human well-being in water research and watershed management. WIREs Water 2,
345–358. https://doi.org/10.1002/wat2.1081.
Burdon, F.J., Ramberg, E., Sargac, J., Forio, M.A., de Saeyer, N., Mutinova, P.T., Moe, T.
F., Pavelescu, M.O., Dinu, V., Cazacu, C., Witing, F., Kupilas, B., Grandin, U.,
Volk, M., Rîs¸noveanu, G., Goethals, P., Friberg, N., Johnson, R.K., McKie, B.G., 2020.
Assessing the benets of forested riparian zones: a qualitative index of riparian
integrity is positively associated with ecological status in european streams. Water.
https://doi.org/10.3390/w12041178.
Campagne, C.S., Roche, P., Müller, F., Burkhard, B., 2020. Ten years of ecosystem
services matrix: Review of a (r)evolution. One Ecosyst. 5, e51103.
Chilkoti, V., Bolisetti, T., Balachandar, R., 2018. Multi-objective autocalibration of SWAT
model for improved low ow performance for a small snowfed catchment. Hydrol.
Sci. J. 63, 1482–1501. https://doi.org/10.1080/02626667.2018.1505047.
European Commission, 2011. Strategy to 2020. Eu. https://doi.org/10.2779/39229.
European Commission, 2020. EU Biodiversity Strategy for 2030 - Bringing nature back
into our lives. COM/2020/380. Brussels.
Cong, W., Sun, X., Guo, H., Shan, R., 2020. Comparison of the SWAT and InVEST models
to determine hydrological ecosystem service spatial patterns, priorities and trade-
offs in a complex basin. Ecol. Ind. 112, 106089 https://doi.org/10.1016/j.
ecolind.2020.106089.
Cosgrove, W.J., Loucks, D.P., 2015. Water management: Current and future challenges
and research directions. Water Resour. Res. 51, 4823–4839. https://doi.org/
10.1002/2014WR016869.
Cs´
aki, P., Gyim´
othy, K., Kalicz, P., Szolgay, J., Zagyvai-Kiss, K.A., Gribovszki, Z., 2020.
Multi-model climatic water balance prediction in the Zala River Basin (Hungary)
based on a modied Budyko framework. J. Hydrol. Hydromechanics 68, 200–210.
https://doi.org/10.2478/johh-2020-0016.
Czúcz, B., Kal´
oczkai, ´
A., Arany, I., Kelemen, K., Papp, J., Havadt˝
oi, K., Campbell, K.,
Kelemen, M.A., V´
ari, ´
A., 2018. How to design a transdisciplinary regional ecosystem
service assessment: A case study from Romania, Eastern Europe. One Ecosyst.
https://doi.org/10.3897/oneeco.3.e26363.
Czúcz, B., Haines-Young, R., Kiss, M., Bereczki, K., Kert´
esz, M., V´
ari, ´
A., Potschin-
Young, M., Arany, I., 2020. Ecosystem service indicators along the cascade: How do
assessment and mapping studies position their indicators? Ecol. Ind. 118, 106729
https://doi.org/10.1016/j.ecolind.2020.106729.
D’Alberto, L., Lucianetti, G., 2019. Misinterpretation of the Kenessey method for the
determination of the runoff coefcient: a review. Hydrol. Sci. J. 64, 288–296.
https://doi.org/10.1080/02626667.2019.1578965.
Decsi, B., V´
ari, ´
A., Kozma, Z., 2020. The effect of future land use changes on hydrologic
ecosystem services: a case study from the Zala catchment, Hungary. Biol. Futur.
https://doi.org/10.1007/s42977-020-00032-6.
Dennedy-Frank, P.J., Muenich, R.L., Chaubey, I., Ziv, G., 2016. Comparing two tools for
ecosystem service assessments regarding water resources decisions. J. Environ.
Manage. 177, 331–340. https://doi.org/10.1016/j.jenvman.2016.03.012.
Díaz, S., Pascual, U., Stenseke, M., Martín-L´
opez, B., Watson, R.T., Moln´
ar, Z., Hill, R.,
Chan, K.M.A., Baste, I.A., Brauman, K.A., Polasky, S., Church, A., Lonsdale, M.,
Larigauderie, A., Leadley, P.W., van Oudenhoven, A.P.E., van der Plaat, F.,
Schr¨
oter, M., Lavorel, S., Aumeeruddy-Thomas, Y., Bukvareva, E., Davies, K.,
Demissew, S., Erpul, G., Failler, P., Guerra, C.A., Hewitt, C.L., Keune, H., Lindley, S.,
Shirayama, Y., 2018. Assessing nature’s contributions to people. Science (80-) 359,
270–272. https://doi.org/10.1126/science.aap8826.
Ding, J., Jiang, Y., Fu, L., Liu, Q., Peng, Q., Kang, M., 2015. Impacts of land use on
surface water quality in a subtropical river basin: a case study of the Dongjiang River
Basin, Southeastern China. Water. https://doi.org/10.3390/w7084427.
Dobor, L., Barcza, Z., Hl´
asny, T., Havasi, ´
A., Horv´
ath, F., Ittz´
es, P., Bartholy, J., 2015.
Bridging the gap between climate models and impact studies: the FORESEE
Database. Geosci. Data J. 2, 1–11. https://doi.org/10.1002/gdj3.22.
European Environment Agency, 2017. CLC2018 Technical Guidelines.
Forio, M.A.E., Villa-Cox, G., Van Echelpoel, W., Ryckebusch, H., Lock, K., Spanoghe, P.,
Deknock, A., De Troyer, N., Nolivos-Alvarez, I., Dominguez-Granda, L., Speelman, S.,
Goethals, P.L.M., 2020. Bayesian Belief Network models as trade-off tools of
ecosystem services in the Guayas River Basin in Ecuador. Ecosyst. Serv. 44, 101124
https://doi.org/10.1016/j.ecoser.2020.101124.
Froese, R., Schilling, J., 2019. The nexus of climate change, land use, and conicts. Curr.
Clim. Chang. Reports 5, 24–35. https://doi.org/10.1007/s40641-019-00122-1.
Gao, J., Li, F., Gao, H., Zhou, C., Zhang, X., 2017. The impact of land-use change on
water-related ecosystem services: a study of the Guishui River Basin, Beijing. China.
J. Clean. Prod. 163, S148–S155. https://doi.org/10.1016/j.jclepro.2016.01.049.
Gassman, W., Reyes, P.R., Green, H.C., Arnold, J., 2007. The soil and water assessment
tool: historical development, applications, and future research directions. Trans.
ASABE 50, 1211–1250. https://doi.org/10.13031/2013.23637.
G´
omez-Baggethun, E., Barton, D.N., Berry, P., Dunford, R., Harrison, P.A., 2016.
Concepts and methods in ecosystem services valuation. Routledge Handb. Ecosyst.
Serv. 99–111.
Grˆ
et-Regamey, A., Sir´
en, E., Brunner, S.H., Weibel, B., 2017. Review of decision support
tools to operationalize the ecosystem services concept. Ecosyst. Serv. 26, 306–315.
https://doi.org/10.1016/j.ecoser.2016.10.012.
Haddaway, N.R., Brown, C., Eales, J., Eggers, S., Josefsson, J., Kronvang, B., Randall, N.
P., Uusi-K¨
ampp¨
a, J., 2018. The multifunctional roles of vegetated strips around and
within agricultural elds. Environ. Evid. 7, 14. https://doi.org/10.1186/s13750-
018-0126-2.
Hadjisolomou, E., Stefanidis, K., Herodotou, H., Michaelides, M., Papatheodorou, G.,
Papastergiadou, E., 2021. Modelling freshwater eutrophication with limited
limnological data using articial neural networks. Water. https://doi.org/10.3390/
w13111590.
Haines-Young, R., Potschin, M.B., 2018. Common international classication of
ecosystem services (CICES) V5. 1 and guidance on the application of the revised
structure.
Halbe, J., Pahl-Wostl, C., Adamowski, J., 2018. A methodological framework to support
the initiation, design and institutionalization of participatory modeling processes in
water resources management. J. Hydrol. 556, 701–716. https://doi.org/10.1016/j.
jhydrol.2017.09.024.
Hamel, P., Bryant, B.P., 2017. Uncertainty assessment in ecosystem services analyses:
Seven challenges and practical responses. Ecosyst. Serv. 24, 1–15. https://doi.org/
10.1016/j.ecoser.2016.12.008.
Hanna, D.E.L., Tomscha, S.A., Ouellet Dallaire, C., Bennett, E.M., 2018. A review of
riverine ecosystem service quantication: research gaps and recommendations.
J. Appl. Ecol. 55, 1299–1311. https://doi.org/10.1111/1365-2664.13045.
Harrison, P.A., Dunford, R., Barton, D.N., Kelemen, E., Martín-L´
opez, B., Norton, L.,
Termansen, M., Saarikoski, H., Hendriks, K., G´
omez-Baggethun, E., Czúcz, B.,
García-Llorente, M., Howard, D., Jacobs, S., Karlsen, M., Kopperoinen, L.,
Madsen, A., Rusch, G., van Eupen, M., Verweij, P., Smith, R., Tuomasjukka, D.,
Zulian, G., 2018. Selecting methods for ecosystem service assessment: a decision tree
approach. Ecosyst. Serv. 29, 481–498. https://doi.org/10.1016/j.
ecoser.2017.09.016.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
13
Harrison-Atlas, D., Theobald, D.M., Goldstein, J.H., 2016. A systematic review of
approaches to quantify hydrologic ecosystem services to inform decision-making.
Int. J. Biodivers. Sci. Ecosyst. Serv. Manag. 12, 160–171. https://doi.org/10.1080/
21513732.2016.1181105.
Hatvani, I.G., de Barros, V.D., Tanos, P., Kov´
acs, J., Sz´
ekely Kov´
acs, I., Clement, A., 2020.
Spatiotemporal changes and drivers of trophic status over three decades in the
largest shallow lake in Central Europe, Lake Balaton. Ecol. Eng. 151, 105861 https://
doi.org/10.1016/j.ecoleng.2020.105861.
Hawkins, R.H., Theurer, F.D., Rezaeianzadeh, M., 2019. Understanding the basis of the
curve number method for watershed models and TMDLs. J. Hydrol. Eng. 24,
06019003. https://doi.org/10.1061/(ASCE)HE.1943-5584.0001755.
Herodek, S., 1986. Phytoplankton changes during eutrophication and P and N
metabolism BT – modeling and managing shallow lake eutrophication. In:
Somly´
ody, L., van Straten, G. (Eds.), Springer. Berlin Heidelberg, Berlin, Heidelberg,
pp. 183–203.
Honti, M., Gao, C., Istv´
anovics, V., Clement, A., 2020. Lessons Learnt from the Long-
Term Management of a Large (Re)constructed Wetland, the Kis-Balaton Protection
System (Hungary). Water 12, 659. https://doi.org/10.3390/w12030659.
Hopkins, K.G., Noe, G.B., Franco, F., Pindilli, E.J., Gordon, S., Metes, M.J., Claggett, P.R.,
Gellis, A.C., Hupp, C.R., Hogan, D.M., 2018. A method to quantify and value
oodplain sediment and nutrient retention ecosystem services. J. Environ. Manage.
220, 65–76. https://doi.org/10.1016/j.jenvman.2018.05.013.
Hu, T., Wu, J., Li, W., 2019. Assessing relationships of ecosystem services on multi-scale:
A case study of soil erosion control and water yield in the Pearl River Delta. Ecol.
Ind. 99, 193–202. https://doi.org/10.1016/j.ecolind.2018.11.066.
Istv´
anovics, V., Clement, A., Somly´
ody, L., Speczi´
ar, A., T´
oth, G., Padis´
ak, J., 2007.
Updating water quality targets for shallow Lake Balaton (Hungary), recovering from
eutrophication. Hydrobiologia 581, 305–318. https://doi.org/10.1007/s10750-006-
0509-1.
Jackson, B., Pagella, T., Sinclair, F., Orellana, B., Henshaw, A., Reynolds, B.,
Mcintyre, N., Wheater, H., Eycott, A., 2013. Polyscape: A GIS mapping framework
providing efcient and spatially explicit landscape-scale valuation of multiple
ecosystem services. Landsc. Urban Plan. 112, 74–88. https://doi.org/10.1016/j.
landurbplan.2012.12.014.
Jacobs, S., Burkhard, B., Van Daele, T., Staes, J., Schneiders, A., 2015. ‘The Matrix
Reloaded’: A review of expert knowledge use for mapping ecosystem services. Ecol.
Model. 295, 21–30. https://doi.org/10.1016/j.ecolmodel.2014.08.024.
Jin, G., Deng, X., Hasan, S.S., Zhao, C., Gibson, J., 2018. Hydrological Ecosystem Services
for Integrated Water Resources Management, in: Deng, X., Gibson, J. (Eds.), Springer
Singapore, Singapore, pp. 1–27. https://doi.org/10.1007/978-981-10-0841-2_6-1.
Jol´
ankai, Z., Kardos, M.K., Clement, A., 2020. Modication of the MONERIS nutrient
emission model for a lowland country (Hungary) to support river basin management
planning in the Danube River Basin. Water. https://doi.org/10.3390/w12030859.
Kandziora, M., Burkhard, B., Müller, F., 2013. Interactions of ecosystem properties,
ecosystem integrity and ecosystem service indicators—A theoretical matrix exercise.
Ecol. Ind. 28, 54–78. https://doi.org/10.1016/j.ecolind.2012.09.006.
Kardos, M.K., Clement, A., 2020. Predicting small water courses’ physico-chemical status
from watershed characteristics with two multivariate statistical methods. Open
Geosci. 12, 71–84. https://doi.org/doi:10.1515/geo-2020-0006.
Keeler, B.L., Polasky, S., Brauman, K.A., Johnson, K.A., Finlay, J.C., O’Neill, A.,
Kovacs, K., Dalzell, B., 2012. Linking water quality and well-being for improved
assessment and valuation of ecosystem services. Proc. Natl. Acad. Sci. 109,
18619–18624. https://doi.org/10.1073/pnas.1215991109.
Koch, J., Demirel, M.C., Stisen, S., 2018. The SPAtial EFciency metric (SPAEF):
multiple-component evaluation of spatial patterns for optimization of hydrological
models. Geosci. Model Dev. 11, 1873–1886. https://doi.org/10.5194/gmd-11-1873-
2018.
Kouadri, S., Elbeltagi, A., Islam, A.R.M.T., Kateb, S., 2021. Performance of machine
learning methods in predicting water quality index based on irregular data set:
application on Illizi region (Algerian southeast). Appl. Water Sci. 11, 190. https://
doi.org/10.1007/s13201-021-01528-9.
Kov´
acs-Hosty´
anszki, A., Bereczki, K., Czúcz, B., Fab´
ok, V., Fodor, L., Kal´
oczkai, ´
A., Kiss,
M., Koncz, P., Kov´
acs, E., Rezneki, R., Tan´
acs, E., T¨
or¨
ok, K., V´
ari, ´
A., Z¨
olei, A.,
Zsembery, Z., 2019. Mapping and assessment of national ecosystem services, or a
national program for nature conservation. “Nemzeti ¨
okosziszt´
ema-szolg´
altat´
as
t´
erk´
epez´
es ´
es ´
ert´
ekel´
es, avagy a term´
eszetv´
edelem orsz´
agos programja.”
Term´
eszetv´
edelmi K¨
ozlem´
enyek 25, 80–90. https://doi.org/10.20332/tvk-
jnatconserv.2019.25.80.
Kozma, Z., Jol´
ankai, Z., Kardos, M.K., Muzel´
ak, B., Koncsos, L., 2022. Adaptive water
management-land use practice for improving ecosystem services – a Hungarian
Modelling Case Study. Period. Polytech. Civ. Eng. 66, 256–268. https://doi.org/
10.3311/PPci.18369.
Kuhnert, M., Voinov, A., Seppelt, R., 2005. Comparing raster map comparison algorithms
for spatial modeling and analysis. Photogramm. Eng. Remote Sens. 71, 975–984.
Landuyt, D., Broekx, S., D’hondt, R., Engelen, G., Aertsens, J., Goethals, P.L.M., 2013.
A review of Bayesian belief networks in ecosystem service modelling. Environ.
Model. Softw. 46, 1–11. https://doi.org/10.1016/j.envsoft.2013.03.011.
Landuyt, D., Broekx, S., Goethals, P.L.M., 2016. Bayesian belief networks to analyse
trade-offs among ecosystem services at the regional scale. Ecol. Ind. 71, 327–335.
https://doi.org/10.1016/j.ecolind.2016.07.015.
Lee, Y., Brody, S.D., 2018. Examining the impact of land use on ood losses in Seoul,
Korea. Land Use Policy 70, 500–509. https://doi.org/10.1016/j.
landusepol.2017.11.019.
Lemm, J.U., Venohr, M., Globevnik, L., Stefanidis, K., Panagopoulos, Y., van Gils, J.,
Posthuma, L., Kristensen, P., Feld, C.K., Mahnkopf, J., Hering, D., Birk, S., 2021.
Multiple stressors determine river ecological status at the European scale: Towards
an integrated understanding of river status deterioration. Glob. Chang. Biol. 27,
1962–1975. https://doi.org/https://doi.org/10.1111/gcb.15504.
Li, Y., Zhang, L., Yan, J., Wang, P., Hu, N., Cheng, W., Fu, B., 2017. Mapping the hotspots
and coldspots of ecosystem services in conservation priority setting. J. Geogr. Sci. 27,
681–696. https://doi.org/10.1007/s11442-017-1400-x.
Liang, J., Li, S., Li, X., Li, X., Liu, Q., Meng, Q., Lin, A., Li, J., 2021. Trade-off analyses and
optimization of water-related ecosystem services (WRESs) based on land use change
in a typical agricultural watershed, southern China. J. Clean. Prod. 279, 123851
https://doi.org/10.1016/j.jclepro.2020.123851.
Liu, Y., De Smedt, F., 2004. A GIS–based Hydrologic Model for Flood Prediction and
Watershed Management. Doc. user manual. Dep. Hydrol. Hydarulic Eng. Vrije Univ.
Brussel 315.
Lüke, A., Hack, J., 2017. Modelling hydrological ecosystem services – A state of the art
model comparison. Hydrol. Earth Syst. Sci. Discuss. 1–29 https://doi.org/10.5194/
hess-2017-436.
Lüke, A., Hack, J., 2018. Comparing the applicability of commonly used hydrological
ecosystem services models for integrated decision-support. Sustain. https://doi.org/
10.3390/su10020346.
Maes, J., Teller, A., Erhard, M., Liquete, C., Braat, L., Berry, P., Egoh, B., Puydarrieux, P.,
Fiorina, C., Santos, F., 2013. Mapping and Assessment of Ecosystems and their
Services. An Anal. Framew. Ecosyst. assessments under action 5, 1–58.
Mahmoud, S.H., Mohammad, F.S., Alazba, A.A., 2014. Determination of potential runoff
coefcient for Al-Baha Region, Saudi Arabia using GIS. Arab. J. Geosci. 7,
2041–2057. https://doi.org/10.1007/s12517-014-1303-4.
Mak´
o, A., T´
oth, B., Hern´
adi, H., Farkas, C., Marth, P., 2010. Introduction of the
Hungarian Detailed Soil Hydrophysical Database (MARTHA) and its use to test
external pedotransfer functions. Agrok´
emia ´
es Talajt. 59, 29–38. https://doi.org/
10.1556/agrokem.59.2010.1.4.
Mengist, W., Soromessa, T., Feyisa, G.L., 2020. A global view of regulatory ecosystem
services: existed knowledge, trends, and research gaps. Ecol. Process. 9, 40. https://
doi.org/10.1186/s13717-020-00241-w.
Nedkov, S., Burkhard, B., 2012. Flood regulating ecosystem services—Mapping supply
and demand, in the Etropole municipality. Bulgaria. Ecol. Indic. 21, 67–79. https://
doi.org/10.1016/j.ecolind.2011.06.022.
Neitsch, S.., Arnold, J.., Kiniry, J.., Williams, J.., 2011. Soil & Water Assessment Tool
Theoretical Documentation Version 2009. Texas Water Resour. Inst. https://doi.org/
10.1016/j.scitotenv.2015.11.063.
Ochoa, V., Urbina-Cardona, N., 2017. Tools for spatially modeling ecosystem services:
Publication trends, conceptual reections and future challenges. Ecosyst. Serv. 26,
155–169. https://doi.org/10.1016/J.ECOSER.2017.06.011.
P´
asztor, L., Laborczi, A., Tak´
acs, K., Szatm´
ari, G., Bakacsi, Z., Szab´
o, J., Ill´
es, G., 2017.
DOSoReMI as the national implementation of GlobalSoilMap for the territory of
Hungary, in: Arrouay, D., Savin, I., Leenaars, J., McBratney, A.B. (Eds.), Proceedings
of the Global Soil Map 2017 Conference. CRC Press, pp. 17–22.
Pham, H.V., Sperotto, A., Furlan, E., Torresan, S., Marcomini, A., Critto, A., 2021.
Integrating Bayesian Networks into ecosystem services assessment to support water
management at the river basin scale. Ecosyst. Serv. 50, 101300 https://doi.org/
10.1016/j.ecoser.2021.101300.
Pinke, Z., Decsi, B., Kozma, Z., V´
ari, ´
A., L¨
ovei, G.L., 2020. A spatially explicit analysis of
wheat and maize yield sensitivity to changing groundwater levels in Hungary,
1961–2010. Sci. Total Environ. 715, 136555 https://doi.org/10.1016/j.
scitotenv.2020.136555.
Pomogyi, P., 1993. Nutrient retention by the Kis-Balaton Water Protection System.
Hydrobiologia. https://doi.org/10.1007/BF00007190.
Qiu, J., Turner, M.G., 2013. Spatial interactions among ecosystem services in an
urbanizing agricultural watershed. Proc. Natl. Acad. Sci. 110, 12149 LP – 12154.
https://doi.org/10.1073/pnas.1310539110.
Redhead, J.W., May, L., Oliver, T.H., Hamel, P., Sharp, R., Bullock, J.M., 2018. National
scale evaluation of the InVEST nutrient retention model in the United Kingdom. Sci.
Total Environ. 610–611, 666–677. https://doi.org/10.1016/j.
scitotenv.2017.08.092.
Rocchini, D., Wegmann, M., Leutner, B., Bevanda, M., 2016. 12 Spatial Land Cover
Pattern Analysis. Remote Sens. GIS Ecol. Using Open Source Softw.
Roche, P.K., Campagne, C.S., 2019. Are expert-based ecosystem services scores related to
biophysical quantitative estimates? Ecol. Ind. 106, 105421 https://doi.org/10.1016/
j.ecolind.2019.05.052.
Rogger, M., Agnoletti, M., Alaoui, A., Bathurst, J.C., Bodner, G., Borga, M., Chaplot, V.,
Gallart, F., Glatzel, G., Hall, J., Holden, J., Holko, L., Horn, R., Kiss, A., Kohnov´
a, S.,
Leitinger, G., Lennartz, B., Parajka, J., Perdig˜
ao, R., Peth, S., Plavcov´
a, L.,
Quinton, J.N., Robinson, M., Salinas, J.L., Santoro, A., Szolgay, J., Tron, S., van den
Akker, J.J.H., Viglione, A., Bl¨
oschl, G., 2017. Land use change impacts on oods at
the catchment scale: Challenges and opportunities for future research. Water Resour.
Res. 53, 5209–5219. https://doi.org/10.1002/2017WR020723.
Rudke, A.P., Xavier, A.C.F., Fujita, T., Abou Rafee, S.A., Martins, L.D., Morais, M.V.B., de
A. Albuquerque, T.T., Freitas, E.D., Martins, J.A., 2021. Mapping past landscapes
using landsat data: Upper Paran´
a River Basin in 1985. Remote Sens. Appl.: Soc.
Environ. 21, 100436 https://doi.org/10.1016/j.rsase.2020.100436.
Sahle, M., Saito, O., Fürst, C., Yeshitela, K., 2019. Quantifying and mapping of water-
related ecosystem services for enhancing the security of the food-water-energy nexus
in tropical data–sparse catchment. Sci. Total Environ. 646, 573–586. https://doi.
org/10.1016/j.scitotenv.2018.07.347.
Sch¨
agner, J.P., Brander, L., Maes, J., Hartje, V., 2013. Mapping ecosystem services’
values: Current practice and future prospects. Ecosyst. Serv. 4, 33–46. https://doi.
org/10.1016/j.ecoser.2013.02.003.
B. Decsi et al.
Ecological Indicators 141 (2022) 109143
14
Schlutow, A., Schr¨
oder, W., 2021. Rule-based classication and mapping of ecosystem
services with data on the integrity of forest ecosystems. Environ. Sci. Eur. 33, 50.
https://doi.org/10.1186/s12302-021-00481-3.
Schmalz, B., Kruse, M., Kiesel, J., Müller, F., Fohrer, N., 2016. Water-related ecosystem
services in Western Siberian lowland basins—Analysing and mapping spatial and
seasonal effects on regulating services based on ecohydrological modelling results.
Ecol. Ind. 71, 55–65. https://doi.org/10.1016/j.ecolind.2016.06.050.
Schr¨
oter, M., Remme, R.P., Sumarga, E., Barton, D.N., Hein, L., 2015. Lessons learned for
spatial modelling of ecosystem services in support of ecosystem accounting. Ecosyst.
Serv. 13, 64–69. https://doi.org/10.1016/j.ecoser.2014.07.003.
Schulp, C.J.E., Burkhard, B., Maes, J., Van Vliet, J., Verburg, P.H., 2014. Uncertainties in
ecosystem service maps: a comparison on the European scale. PLoS ONE 9, 1–11.
https://doi.org/10.1371/journal.pone.0109643.
Seibert, J., Staudinger, M., van Meerveld, H.J., 2019. Validation and over-
parameterization—experiences from hydrological modeling. Springer International
Publishing, Cham, pp. 811–834, 10.1007/978-3-319-70766-2_33.
Shackleton, C.M., Ruwanza, S., Sinasson Sanni, G.K., Bennett, S., De Lacy, P., Modipa, R.,
Mtati, N., Sachikonye, M., Thondhlana, G., 2016. Unpacking Pandora’s Box:
understanding and categorising ecosystem disservices for environmental
management and human wellbeing. Ecosystems 19, 587–600. https://doi.org/
10.1007/s10021-015-9952-z.
Shamshirband, S., Jafari Nodoushan, E., Adolf, J.E., Abdul Manaf, A., Mosavi, A.,
Chau, K., 2019. Ensemble models with uncertainty analysis for multi-day ahead
forecasting of chlorophyll a concentration in coastal waters. Eng. Appl. Comput.
Fluid Mech. 13, 91–101. https://doi.org/10.1080/19942060.2018.1553742.
Sharp, R., Tallis, H.T., Ricketts, T., Guerry, A.D., Wood, S.A., Chaplin-Kramer, R., Nelson,
E., Ennaanay, D., Wolny, S., Olwero, N., Vigerstol, K., Pennington, D., Mendoza, G.,
Aukema, J., Foster, J., Forrest, J., Cameron, D., Arkema, K., Lonsdorf, E., Kennedy,
C., Verutes, G., Kim, C.K., Guannel, G., Papenfus, M., Toft, J., Marsik, M., Bernhardt,
J., Grifn, R., Glowinski, K., Chaumont, N., Perelman, A., Lacayo, M.M.L., Hamel, P.,
Vogl, A.L., Rogers, L., Bierbower, W., Denu, D., Douglass, J., 2018. InVEST 3.7.0
User’s Guide. Nat. Cap. Proj. Stanford Univ. Univ. Minnesota, Nat. Conserv. World
Wildl. Fund.
Souliotis, I., Voulvoulis, N., 2022. Operationalising nature-based solutions for the design
of water management interventions. Nature-Based Solut. 2, 100015 https://doi.org/
10.1016/j.nbsj.2022.100015.
Sta´
nczuk-Gałwiaczek, M., Sobolewska-Mikulska, K., Ritzema, H., van Loon-Steensma, J.
M., 2018. Integration of water management and land consolidation in rural areas to
adapt to climate change: experiences from Poland and the Netherlands. Land use
policy 77, 498–511. https://doi.org/10.1016/j.landusepol.2018.06.005.
Steinhoff-Knopp, B., Kuhn, T.K., Burkhard, B., 2021. The impact of soil erosion on soil-
related ecosystem services: development and testing a scenario-based assessment
approach. Environ. Monit. Assess. 193, 274. https://doi.org/10.1007/s10661-020-
08814-0.
Sun, G., Hallema, D., Asbjornsen, H., 2017. Ecohydrological processes and ecosystem
services in the Anthropocene: a review. Ecol. Process. 6, 35. https://doi.org/
10.1186/s13717-017-0104-6.
Sutherland, I.J., Villamagna, A.M., Dallaire, C.O., Bennett, E.M., Chin, A.T.M., Yeung, A.
C.Y., Lamothe, K.A., Tomscha, S.A., Cormier, R., 2018. Undervalued and under
pressure: A plea for greater attention toward regulating ecosystem services. Ecol.
Ind. 94, 23–32. https://doi.org/10.1016/j.ecolind.2017.06.047.
Tan, M.L., Gassman, P.W., Yang, X., Haywood, J., 2020. A review of SWAT applications,
performance and future needs for simulation of hydro-climatic extremes. Adv. Water
Resour. 143, 103662 https://doi.org/10.1016/j.advwatres.2020.103662.
Tan´
acs, E., Bel´
enyesi, M., Lehoczki, R., Pataki, R., Petrik, O., Standov´
ar, T., P´
asztor, L.,
Laborczi, A., Szatm´
ari, G., Moln´
ar, Z., Bede-Fazekas, ´
A., Somodi, I., Krist´
of, D.,
Kov´
acs-Hosty´
anszki, A., T¨
or¨
ok, K., Kisn´
e Fodor, L., Zsembery, Z., Friedl, Z.,
Maucha, G., 2021. Compiling a high-resolution country-level ecosystem map to
support environmental policy: methodological challenges and solutions from
Hungary. Geocarto Int. 1–24 https://doi.org/10.1080/10106049.2021.2005158.
T´
atrai, I., M´
aty´
as, K., Korponai, J., Paulovits, G., Pomogyi, P., 2000. The role of the Kis-
Balaton Water Protection System in the control of water quality of Lake Balaton.
Ecol. Eng. 16, 73–78. https://doi.org/10.1016/S0925-8574(00)00091-4.
Turkelboom, F., Demeyer, R., Vranken, L., De Becker, P., Raymaekers, F., De Smet, L.,
2021. How does a nature-based solution for ood control compare to a technical
solution? Case study evidence from Belgium. Ambio 50, 1431–1445. https://doi.
org/10.1007/s13280-021-01548-4.
Turunen, J., Elbrecht, V., Steinke, D., Aroviita, J., 2021. Riparian forests can mitigate
warming and ecological degradation of agricultural headwater streams. Freshw.
Biol. 66, 785–798. https://doi.org/10.1111/fwb.13678.
Vallecillo, S., Kakoulaki, G., La Notte, A., Feyen, L., Dottori, F., Maes, J., 2020.
Accounting for changes in ood control delivered by ecosystems at the EU level.
Ecosyst. Serv. 44, 101142 https://doi.org/10.1016/j.ecoser.2020.101142.
V´
ari, ´
A., Kozma, Z., Pataki, B., Jol´
ankai, Z., Kardos, M.K., Decsi, B., P´
asztor, L., Bakacsi,
Z., T´
oth, B., Laborczi, A., Pinke, Z., Jol´
ankai, G., Centeri, C., Matt´
anyi, Z., D´
oka, R.,
Kisn´
e Fodor, L., Zsembery, Z., 2021. Assessment of ood risk mitigation in lowland
and hilly areas, erosion control, nutrient ltration and drought mitigation as
ecosystem services - From ecosystem status to evaluation of actually used ecosystem
service. Budapest, Hungary. https://doi.org/10.34811/osz.hidrologia.tanulmany.
V´
ari, ´
A., Podschun, S.A., Er˝
os, T., Hein, T., Pataki, B., Ioj˘
a, I.-C., Adamescu, C.M.,
Gerhardt, A., Gruber, T., Dedi´
c, A., ´
Ciri´
c, M., Gavrilovi´
c, B., B´
aldi, A., 2022.
Freshwater systems and ecosystem services: challenges and chances for cross-
fertilization of disciplines. Ambio 51, 135–151. https://doi.org/10.1007/s13280-
021-01556-4.
Vigerstol, K.L., Aukema, J.E., 2011. A comparison of tools for modeling freshwater
ecosystem services. J. Environ. Manage. 92, 2403–2409. https://doi.org/10.1016/J.
JENVMAN.2011.06.040.
Wang, Z., Lechner, A.M., Baumgartl, T., 2018. Ecosystem services mapping uncertainty
assessment: a case study in the tzroy Basin Mining Region. Water. https://doi.org/
10.3390/w10010088.
Western Transdanubian Directorate of Water Management, 2016. Second River Basin
Management Plan - 4.1. Zala sub-catchment. Szombathely, Hungary.
Wheater, H., Evans, E., 2009. Land use, water management and future ood risk. Land
use policy 26, S251–S264. https://doi.org/10.1016/j.landusepol.2009.08.019.
Whittaker, G., Confesor Jr., R., Di Luzio, M., Arnold, G., 2010. Detection of
overparameterization and overtting in an automatic calibration of SWAT. Trans.
ASABE 53, 1487–1499. https://doi.org/10.13031/2013.34909.
Willcock, S., Martínez-L´
opez, J., Hooftman, D.A.P., Bagstad, K.J., Balbi, S., Marzo, A.,
Prato, C., Sciandrello, S., Signorello, G., Voigt, B., Villa, F., Bullock, J.M.,
Athanasiadis, I.N., 2018. Machine learning for ecosystem services. Ecosyst. Serv. 33,
165–174. https://doi.org/10.1016/j.ecoser.2018.04.004.
Willemen, L., Crossman, N.D., Quatrini, S., Egoh, B., Kalaba, F.K., Mbilinyi, B., de
Groot, R., 2018. Identifying ecosystem service hotspots for targeting land
degradation neutrality investments in south-eastern Africa. J. Arid Environ. 159,
75–86. https://doi.org/10.1016/j.jaridenv.2017.05.009.
Zhang, W., Ricketts, T.H., Kremen, C., Carney, K., Swinton, S.M., 2007. Ecosystem
services and dis-services to agriculture. Ecol. Econ. 64, 253–260. https://doi.org/
10.1016/j.ecolecon.2007.02.024.
B. Decsi et al.
